【作者】 周治国； 李怡瑶； 屈崇； 贾书丽；

【Author】 ZHOU Zhiguo;LI Yiyao;QU Chong;JIA Shuli;School of Information and Electronics,Beijing Institute of Technology;Shanghai Marine Diesel Engine Research Institute;

【机构】 北京理工大学信息与电子学院； 中国船舶重工集团公司第七一一研究所；

【摘要】 船用柴油机监测数据具有样本数据多、特征呈非线性等特点,使得状态预测比较困难。为提高船舶运行可靠性和数据预测的实时性,利用某船用柴油机实际运行数据,提出一种基于LightGBM机器学习算法的柴油机运行状态预测模型,利用箱线图剔除噪声点降低信息损失。预测结果表明,算法模型对两种不同特征参数均保证较高准确率,且构建迅速,具有高可用性和高实时性。更多还原

【Abstract】 The monitoring data of marine diesel engine has the characteristics of many sample data and non-linear characteristics,which makes the state prediction more difficult.In order to improve the operational reliability of the ship and the real-time performance of data prediction,this paper uses the actual operating data of a marine diesel engine to propose a diesel engine operating state prediction model based on the LightGBM machine learning algorithm,and uses the box plot to eliminate noise points to reduce information loss.The prediction results show that the algorithm model guarantees high accuracy for two different characteristic parameters,and is constructed quickly,with high availability and high real-time performance.更多还原